A household's power demand disaggregated into appliances

In order to perform such a disaggregation task, it is necessary to first build a mathematical model of a household. My work has adopted a

hidden Markov model (HMM) based approach, in which appliances within a household are represented by HMMs. Each HMM contains a number of parameters which describe the behaviour of a particular appliance, such as its power demand or frequency of use. Since households contain many appliances, a household model can be constructed using a combination of HMMs (one for each appliance), a model known as the factorial hidden Markov model (FHMM).

A hidden Markov model

There are many well studied algorithms for learning the parameters and also inferring unknown variables in HMMs and FHMMs. However, such approaches typically make assumptions that do not match the real-world scenario, such as requiring sub-metered data from each appliance in the home. This is obviously not a fair assumption, since it violates the 'non-intrusive' requirement for our system. My PhD has so far focused on the development of training and inference algorithms for such HMMs and FHMMs which enable NIALM technologies to be applied in the real world.